Tekla MCP Server

Tekla MCP Server

What is Tekla MCP Server about?

Tekla MCP Server acts as a bridge between AI agents or MCP‑compatible clients and Tekla Structures. It exposes a set of tools and resources that allow users to control modeling tasks—such as selecting objects, inserting components, adjusting properties, and manipulating views—through plain‑text instructions.

How to use Tekla MCP Server?

1. Install Tekla Structures (tested with versions 2022 and 2026).
2. Set up the MCP client according to the Quickstart guide in the repository.
3. Run the server (Python environment, see documentation for the exact command).
4. Send natural‑language requests from an LLM‑powered client (e.g., Claude, Gemini, GPT‑4o). The server interprets the request, maps it to Tekla API calls, and returns a response.

Key features of Tekla MCP Server

• Modular architecture powered by FastMCP 3.0 with providers for Selection, View, Properties, Components, Operations, Drawing.
• Automatic discovery of filters, macros, components, and custom requirements.
• Plugin system for component handlers, enabling lifecycle hooks and custom logic (e.g., Lifting Anchors component).
• LLM‑driven property understanding that maps natural language (e.g., “concrete cover thickness”) to actual Tekla attributes.
• Hybrid semantic attribute mapping using MiniLM with LLM fallback for robust name‑to‑attribute resolution.
• Compatibility with a wide range of language models and MCP clients.

Use cases of Tekla MCP Server

• Rapid design iteration: Engineers dictate changes (“add a 300 mm diameter rebar column at X,Y”) and the server creates the element instantly.
• Quality assurance scripts: Automated checks (“list all beams without fire‑proof coating”) run via natural language.
• Custom component automation: Plug‑in components like anchoring systems compute positions based on part weight or geometry.
• Training and onboarding: New users interact with Tekla through conversational prompts rather than learning the full UI.

FAQ

Q: Which Tekla versions are supported?
A: Tested with Tekla 2022 and Tekla 2026; other versions may work but are not officially verified.

Q: Do I need a specific AI model?
A: Any MCP‑compatible client can be used; the server has been validated with Claude Sonnet 4.6, Gemini Flash, GPT‑4o, and several open‑source models.

Q: How are plugins added?
A: Develop a Python module that follows the component handler interface and register it in the server’s configuration file.

Q: Is the project open‑source?
A: Yes, it is released under GPL‑v3.

Q: Where can I find full documentation?
A: See the docs folder in the repository for Reference, Quickstart, Configuration, Testing, and Distribution guides.

Tekla MCP Server

This server facilitates interaction with Tekla Structures, helping users automate and accelerate modeling workflows. It acts as a bridge between AI agents or MCP-compatible clients and Tekla, exposing resources and tools for selection, component insertion, property management and view operations.

📌 What is MCP?

MCP stands for Model Context Protocol, and it is a communication protocol introduced by Anthropic to enable more efficient and secure interactions between large language models and other systems, such as human users or other AI agents.

Tekla MCP Server uses AI-powered natural language processing to make interactions more intuitive, allowing user to work with tools using plain text.

To use this server, one must first install and configure an MCP client.

Features

• Modular Architecture: Powered by FastMCP 3.0, with toolset organization through modular providers (Selection, View, Properties, Components, Operations, Drawing).

• Resource Discovery: Auto‑detection of available filters, macros, components, and custom requirements and instructions.

• Component Handler Plugin System: Flexible plugin model for Tekla components with lifecycle hooks, e.g., Lifting Anchors component select anchors based on element weight and auto‑calculates anchor placement according to center of gravity.

• LLM‑Powered Component Property Understanding: Natural language mapping like "concrete cover thickness" → actual Tekla component property.

• Semantic Attribute Mapping: Hybrid semantic system (MiniLM + LLM fallback) for mapping user‑friendly names to Tekla attributes.

See Reference for complete list of tools and resources.

Compatibility

The server was tested to work with Tekla 2022 and Tekla 2026 and may not be compatible with other versions of Tekla Structures.

Verified to work correctly with Claude Desktop, DeepChat and Goose clients, along with the following language models:

• Claude Sonnet 4.6
• Claude Haiku 4.5
• DeepSeek-V3
• DeepSeek-V4-Flash
• Gemini 2.0 Flash
• Gemini 2.5 Flash
• Qwen3
• GPT-4o
• gpt-oss

Installation

For complete setup instructions, see Quickstart Guide.

Documentation

License

This software is open-source and released under the GPLv3 license. You are free to use, modify, and distribute it, as long as all modifications remain open-source under the same license.

For full details, please refer to the LICENSE file included in this repository.

Disclaimer

This software is provided as is, without any warranties or guarantees of functionality, reliability, or security. The developer assumes no responsibility for any damages, data loss, or other issues arising from its use.

Use at your own risk.